Notes from the factory floor and firmware repositories.
One of the hardest problems in bare-hand optical tracking is dealing with occlusion—when one finger blocks the sensor's view of another. We discuss how we implemented a predictive kinematic model to maintain cursor stability even when the raw optical data drops out for a few milliseconds.
Read full log →Vertical travel is the enemy of optical tracking stability. Every millimeter your hand moves up and down while typing introduces noise to the baseline spatial data. Here is the math behind why we went exclusively with low-profile switches.
Read full log →The optical array requires absolute rigidity. A chassis flex of even 0.2mm throws off the triangulation math. Notes on refining our aluminum tooling to ensure the sensor mounts remain perfectly static under heavy typing loads.
Read full log →